🎙️ TRANSCRIBE : Audio-to-Text Transcription with Whisper and Tkinter

This user-friendly Python application converts audio files (MP3, WAV, M4A, FLAC, etc.) into text using OpenAI’s Whisper model via a simple, multilingual Tkinter GUI.

---

🚀 Key Features

• 🌐 Multilingual interface:
  • Full support for English and French UI.
  • Over 30 transcription languages available.
  • 🔄 Dynamic language switching: Change interface language live from the dropdown (top left), or during first startup via popup.
• 📁 Multi-file support: Select and transcribe multiple audio files at once.
• 🧠 Smart model selection: Choose from Whisper models: tiny, base, small, medium, large.
• 📦 Auto-saving:
  • Files are saved with timestamped filenames.
  • Results are stored in the output/ folder.
• 🌀 Progress feedback:
  • A spinner animates while transcribing.
  • Status label changes from "Transcribing..." to "Finished.".
• 📂 Output access:
  • Click View transcriptions to open the output folder.
  • Transcriptions are also shown live in the interface.
• 🔧 Persistent preferences:
  • Last selected model and language are saved in a local config.json file.
• ⚙️ Lightweight and easy to install.

---

📋 Prerequisites

• Python ≥ 3.8
• Whisper speech-to-text engine:

pip install openai-whisper

• FFmpeg (required for audio decoding):

Windows

winget install ffmpeg

Or download manually from ffmpeg.org and add the bin folder to your system PATH.

Linux/macOS

sudo apt install ffmpeg  # Debian/Ubuntu
brew install ffmpeg      # macOS (with Homebrew)

---

🛠️ Installation

1. Clone this repository:

git clone https://github.com/Pizzakira/Transcribe
cd transcribe

2. (Optional) Create a virtual environment:

python -m venv .venv
source .venv/bin/activate      # Linux/macOS
.venv\Scripts\activate         # Windows

3. Install dependencies:

pip install openai-whisper

4. Verify FFmpeg is installed:

ffmpeg -version

---

▶️ How to Use

Launch the app

• Windows: Double-click transcribe_gui.py or create a .bat file.
• macOS/Linux:

python transcribe_gui.py

Step-by-step guide

1. Choose file(s): Click the Choose File… button. Multiple file selection is supported.
2. Model: Pick a Whisper model (tiny, base, small, etc.).
3. Transcription language:
   • Choose Automatic or select a specific language.
4. Click Transcribe: Wait while processing.
5. View results:
   • Transcription text appears in the interface.
   • Output is saved in the output/ folder.
6. Change UI language:
   • Use the interface language selector (top-left) to switch between French/English.

---

Roadmap

✅ Phase 1 — Core Features (Completed)

Error recovery
→ Resumes an interrupted transcription session using saved state.
Status: ✅ Implemented

Batch transcription
→ Allows selection of multiple audio files and processes them sequentially.
Status: ✅ Implemented

Persistent configuration
→ Stores user preferences (Whisper model, transcription language) in config.json.
Status: ✅ Implemented

✅ Phase 2 — Technical and Maintenance (Completed)

Memory release (GPU/CPU)
→ Releases resources after transcription to avoid overload.
Status: ✅ Implemented

Model update check
→ Checks for Whisper model updates and notifies the user.
Status: ❌ Dropped (feature removed at your request)

Automatic ffmpeg verification
→ Checks if ffmpeg is available at launch, displays error otherwise.
Status: ✅ Implemented

🟡 Phase 3 — Developer Experience (In Progress)

Docstrings for all functions
→ Improves readability and maintainability.
Status: 🔄 Planned

Type annotations (type hints)
→ Clarifies expected argument and return types.
Status: 🔄 Planned

Modular code structure
→ Separates logic into distinct files: UI, transcription engine, configuration.
Status: 🔄 Planned

---

📖 Contributing

Contributions are welcome! Feel free to open issues or submit pull requests.